JPH1139471A - Segment element extraction system and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate the extent of deterioration of a structure by extracting an image area including segment elements by removing noise from an object to be evaluated while saving a cracking and a chipping area in a picked-up image. SOLUTION: An image of the outward appearance of the structure is picked up and, for example, a binary image is cut out of the image by using a scanning window consisting of n×n pixels, and 1 or 0 is arranged in the figure in the order of a0 , a1 ...a15 to obtain a numeral between 0 and 65535. This numeral is used as an address to output 1 when a 1/0 pattern in the 4×4 window includes a segment element in a memory or 0 when not. Then is decided by scan units of the scanning window whether or not there is a segment element and when it is decided that there is the corresponding segment element, it is stored as secondary image data in the memory together with the position of a pixel of interest. Consequently, only an image including segment elements is extracted as a binary image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image of a structural condition inspection apparatus, an image of a product, a material, and the like, which captures an external image of a structure or an object and processes the signal to extract an external defect such as a crack or a chip. Used for visual inspection equipment by processing.

[0002]

2. Description of the Related Art In the evaluation and inspection of cracks and chips on the surface of an object as described above, an image of the surface of the object is taken by a TV camera or a line camera as shown in FIG. 1 to evaluate the surface image. . However, the surface of the object is a screen having a large number of irregularities such as a concrete wall surface, and the captured image is an image in which cracks and chips are buried in a large number of minute point-like noises as shown in FIG. This amounts to several thousand points, and it is very difficult to individually analyze and extract a linear area in a short time.

[0003] Examples of developments for similar objects include "Development of Tunneling Laser Measurement System" described on pages 14 to 18 of the May 1997 issue of "Construction Machinery" magazine. An effective method for eliminating minute dot-like noise while reliably extracting is not specified.

[0004]

SUMMARY OF THE INVENTION The problem to be solved by the present invention is as follows.
The purpose is to extract cracks or chips to be detected from among many minute point-like noises.

[0005]

For this reason, a method is provided for removing noise while preserving cracks and missing areas in a captured image. Here, a pixel data group forming an image is converted into a scanning window including one target pixel and at least one reference pixel adjacent to the target pixel in accordance with the arrangement order when the image is drawn with the image data. Is sequentially scanned, and for each scanning unit, the presence or absence of a line segment element extending to the reference pixel through the pixel of interest is determined based on data of the pixel of interest and the reference pixel, and it is determined that there is a corresponding line segment element. At times, a method of extracting line segment elements from image data by storing and accumulating this as secondary image data together with the position data of the pixel of interest in the scanning unit at that time is used.

As a scanning window, specifically, a scanning window of n pixels × n pixels is used. As shown in FIG. 3, a window of n pixels × n pixels is scanned on a binary image including a defective area. I do.

In addition, the presence / absence of a line segment element extending through a pixel of interest and extending to a reference pixel, which is performed for each scanning unit of a scan window, is determined by determining whether a line element extending through a pixel of interest and a reference pixel adjacent thereto is extended. Judge by whether there is.

In this regard, a scanning window of 4 pixels × 4 pixels is used, a pixel corresponding to the second row from the top of the window and the third column from the left is set as a target pixel, and the other 15 pixels are used.
The case where the number of pixels is a reference pixel will be described. In a scan window of 4 pixels × 4 pixels, a line segment passing through the window of interest and the reference pixel and crossing the window has a pattern group as shown in FIG. 4, for example. In FIG. 4, in the vertical or horizontal direction, in each row and column, the pixels adjacent in the horizontal, vertical, and diagonal directions are 1 (dark), and when it can be determined as one element of the line segment, the pixel indicated by * Whatever value is taken, the target pixel of the net pattern is left as a line segment element pixel in the processing result image of FIG.

The determination of the presence or absence of a line segment element extending to the reference pixel adjacent to the pixel passing through the target pixel is performed by, for example, a template reference method or a table lookup method.

In the template reference method, all the pattern of line segment elements extending to the reference pixel through the pixel of interest are stored, and for each scanning unit of the scanning window, the detected pattern corresponds to the template pattern related to the storage. Check whether or not.

In an example of a scanning window of 4 pixels × 4 pixels, there are 48 patterns in the basic pattern group as shown in FIG. 4, in which the target pixel is always “1” (dark). Therefore, assuming that the * part may be either 1 (dark) or 0 (light), the total number of patterns that can be used as line segment elements is
There are more than 20,000. Therefore, to execute this in the template reference method, it is necessary to prepare a large number of template patterns and compare them with all the patterns.
Requires a relatively long processing time.

The table lookup system can determine whether there is a line segment element in a shorter processing time than the template reference system. The table look-up method will be described with reference to a 4.times.4 scanning window. A binary image is cut out from the image in a 4 × 4 window shown in FIG. In the figure, 1 or 0 is arranged in the order of A ₀ , A ₁ ,..., A ₁₅ to obtain a 16-bit numerical value.
Numerical values up to 5. Using this numerical value as an address, 64
The circuit of FIG. 6 is constructed so as to refer to a kilobit memory, and in the memory, one of the 4 × 4 windows is stored.
When the / 0 pattern includes a line segment element, 1 is output, and when the / 0 pattern does not include a line element, 0 is output.

As described above, the presence / absence of a line segment passing through the pixel of interest and the reference pixel in the scanning window is determined for each scanning unit of the scanning window, and when it is determined that the corresponding line segment exists, , Together with the position of the pixel of interest as secondary image data in the memory. As a result, only the image including the line segment element is extracted as a binary image.

[0014]

FIG. 1 shows an embodiment of the present invention. FIG. 1 shows an apparatus that captures an external appearance image of a structure, detects cracks and chips, and measures the size and number of the structures. FIG. 7 shows the overall flow of the processing. From the multi-valued image input by the image input device of FIG. 1, a portion that is darker than the surroundings is detected by a brightness difference detection filter as shown in FIG. In the brightness difference detection filter, a brightness difference Δf ₀ from a pixel separated by L pixels in the vertical, horizontal, and oblique directions from the target image is calculated. It is calculated by the following formula and placed at a position corresponding to the target pixel of the output image. Δf ₀ = (f ₁ + f ₂ + f ₃ + f ₄ + f ₅ + f ₆ + f ₇
+ F ₈ ) -f ₀

In this output image, since Δf ₀ has a larger value in a darker place than its surroundings, as shown in the flowchart of FIG. 7, when Δf ₀ is larger than an appropriate threshold value Th, the corresponding image Defects such as cracks and chips can be extracted by binarizing the image by setting the position to 1 and the others to 0.
As shown in the figure, a large number of fine noise components remain due to unevenness of the surface other than the defective portion.

Here, by operating the line segment element extraction circuit of the present invention, only the line segment elements can be left, but a cluster area larger than 4 pixels also remains. By the way, the line segment element extraction circuit shown in FIG. 6 shortens the length of the line segment by 3 pixels at both ends, and also reduces the surrounding pixels by 3 pixels at both ends each time in the lump area. Therefore, when the element extraction circuit for n lines is operated, the width becomes 3n.
The area within the pixel disappears. However, if it is a line segment element, it will not be erased if its length is 3n pixels or more. In this way, only the line segment elements can be left.

The remaining area is only the area including the line segment element, and the number of areas can be reduced from several tenths to several hundredths of the number of areas immediately after binarization. The region is extracted by a method such as normal labeling, the position is determined, and as shown in FIG. 7, returning to the image immediately after the binarization, only the region including the region is left from the binarized image. I do. Since most of the remaining area is a line segment area, its width and length are measured and used for inspection. Here, as a result of measuring the width and length at this stage, if the width or length is not cracked or chipped, it is excluded from the measurement target.

[0018]

As described above, the degree of deterioration of a structure can be estimated by evaluating the degree of cracking or chipping.

[Brief description of the drawings]

FIG. 1 is an example of a structure appearance imaging device.

FIG. 2 is an example of a structure image including cracks and chips.

FIG. 3 shows a method of scanning a window in an image.

FIG. 4 is an example of a part of 48 basic patterns for extracting line segment elements.

FIG. 5 shows a concept of converting a 1/0 value in a window of a binary image into an address value.

FIG. 6 is an example of a line segment element extraction circuit using a table lookup method.

FIG. 7 shows a flow of processing of the entire inspection apparatus.

FIG. 8 is an explanatory diagram of a brightness difference detection filter from the surroundings.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Inspection object structure 2 Line camera 3 Lighting 4 Traveling vehicle 5 Image recording device 6 Image analysis device

Claims (4)

[Claims] 1. An image evaluation / inspection apparatus comprising an image pickup circuit and an image processing circuit, wherein a defective portion is extracted from an image having many noise components by extracting only an image region including a line segment element from an evaluation target. , And its equipment. 2. A line segment element extraction method and apparatus using the structure and method described in claim 1, wherein a line segment may be formed in an n × n window scanned on a plane of a binary image. By storing all the patterns as a template and leaving only the pixels containing that pattern,
A method for extracting a line segment region, and a device therefor. 3. A line segment element extraction method and apparatus using the configuration and method described in claim 2, wherein a binary image value in an n × n window is set as an n ² bit address value, A method of extracting a line segment area in which a line element can be extracted at high speed by referring to the prepared data in the memory and recording whether or not the data is a line element, and the method. apparatus 4. A pixel data group constituting an image is classified into one attention pixel and at least one pixel adjacent to the attention pixel in accordance with the arrangement order when the image is drawn with the image data.
Scanning is sequentially performed in a scanning window including two reference pixels, and for each scanning unit, the presence or absence of a line segment element extending to the reference pixel through the pixel of interest is determined based on data of the pixel of interest and the reference pixel. When it is determined that there is a line segment element, the line segment element extraction method for extracting the line segment element from the image data by storing and storing this as secondary image data together with the position data of the target pixel in the scanning unit at that time, and its apparatus.